In recent years, even though the quantity of information (transmission capacity) transmittable over one optical fiber continues to increase as the number of multiplexed wavelength channels increases and the modulation speed of an optical signal becomes faster and faster. The increase of the transmission capacity reaches a limit of approximately 10 T (Tera)bit/s. The main reason why it is difficult to increase the transmission capacity is that a wavelength bandwidth usable for the optical transmission reaches the maximum bandwidth of a wavelength bandwidth (a sum of C-band, L-band, and S-band corresponds to approximately 80 nm=10 THz) of an optical fiber amplifier. In order to further increase the transmission capacity of the optical fiber, a study was performed on a signal modulation scheme to increase the efficiency of spectral usage by packing as much as possible a number of optical signals in the limited spectrum.
In the world of wireless communication, the efficiency of spectral usage exceeds 10 by a multilevel modulation technology spread since the 1960s. As a result, high-efficiency transmission has been achieved. Since multilevel modulation has great prospects for the optical fiber transmission, many studies on the multilevel modulation have progressed from the past. For example, R. A. Griffin, et. al., “10 GB/s Optical Differential Quadrature Phase Shift Key (DQPSK) Transmission using GaAs/AlGaAs Integration,” OFC2002, paper PD-FD6, 2003 (Non-Patent Document 1) discloses QPSK (Quadrature Phase Shift Keying) that performs quaternary phase shift keying and Kenro Sekine, Nobuhiko Kikuchi, Shinya Sasaki, Shigenori Hayase and Chie Hasegawa, “Proposal and Demonstration of 10-Gsymbol/sec 16-ary (40 Gbit/s) Optical Modulation/Demodulation Scheme”, paper We3.4.5, ECOC 2004, 2004 (Non-Patent Document 2) discloses 16-level amplitude and phase modulation that is a combination of quaternary amplitude shift keying and quaternary phase shift keying.
In FIG. 1, (A) to (D) show features of various modulation schemes in the prior art applicable to optical transmission, wherein a signal point (complex representation of the optical field at the decision timing) of optical modulation is plotted on a complex plane (IQ plane).
FIG. 1(A) shows a binary amplitude shift keying (BASK) that is widely being used. The BASK transmits 1-bit signal by only using the amplitude (high and low) while not using a phase.
FIG. 1(B) shows a quaternary phase shift keying (QPSK) that transmits 2-bit signal (11, 10, 01, 00) at one symbol by using a quaternary phase angle (0, π/2, π, −π/2).
FIG. 1(C) shows 16-level quadrature amplitude modulation (16QAM) widely being used in wireless communication. The 16QAM, which has signal points arranged in a grid shape, can transmit 4-bit signal at 1 symbol.
In the example shown, a value of the upper 2 bits (11xx, 10xx, 01xx, 00xx) is represented on a Q-axis coordinate and a value of the lower 2 bits (xx11, xx10, xx01, xx00) is represented on an I-axis coordinate. Since the arrangement of the signal points makes a signal point distance large, it has been known that receiver sensitivity is high. However, an implementation example in a field of the optical communication has not yet been reported.
FIG. 1(D) shows a 16-level amplitude and phase modulation (16APSK) where signal points of a binary amplitude shift keying and signal points of a 8-level phase shift keying are arranged in a concentric circular shape and FIG. 1(E) shows a relationship between the amplitude and phase.
As described above, although various arrangements of the signal points of the multilevel signal are reviewed from the past, the receiver becomes complicated as the number of multilevel increases. Further, if the number of multilevel increases, an inter-symbol interference in optical delayed detection for demodulating phase components increases, resulting in a problem that characteristics, such as receiver sensitivity, are rapidly degraded.
On the other hand, in order to increase the optical transmission capacity, a scheme that increases the modulation speed of each wavelength (channel) to about 10 Gbit/s to 40 Gbit/s has been studied. If the modulation speed is increased as described above, however, signal quality is significantly degraded due to the chromatic dispersion in the optical fiber or fiber non-linear effects, such as self-phase modulation. In the case of the optical transmission, the optical transmission distance is rapidly decreased as 1/(signal bit rate)2 due to the influence of the chromatic dispersion. For this reason, in the optical transmission of 10 Gbit/s or more, an optical signal receiving end or an optical repeater should have optical dispersion compensators for compensating the chromatic dispersion in a transmission line. For example, in the optical transmission of 40 Gbit/s, since tolerance against the chromatic dispersion is no more than 5 km for a standard single-mode fiber, an adaptive compensation technology, which automatically controls a tunable chromatic dispersion compensator disposed in the optical signal receiving end so as to minimize the degradation of the signal quality, has been studied.
However, the tunable chromatic dispersion compensator has many problems to be solved, such as the size, complexity, cost, control speed, and the like of the device. In recent years, a configuration that disposes an electrical adaptive equalizer, such as a feed-forward equalizer (FEE) or a decision-feedback equalizer (DFE), in an electrical circuit of the optical signal receiver or an electric stage compensation technology that estimates a receiving symbol using a most likelihood sequence estimator (MLSE) has been studied. However, the chromatic dispersion compensation in the electric stage according to the prior art is incomplete because only an eye opening of a received optical waveform is enlarged. For this reason, the compensation effect is still not sufficient because it can effectively expand the chromatic dispersion tolerance of the receiver to 1.5 to 2 times, for example, and extend the transmission distance of 40 Gbit/s signals to just 10 km in the standard optical fiber transmission.
As one of the prior arts that can solve the above-mentioned problems, for example, there is a coherent optical field receiving system that is disclosed in M. G. Taylor, “Coherent Detection Method Using DSP to Demodulate Signal and for Subsequent Equalization of Propagation Impairments,” paper We4.P.111, ECOC 2003, 2003 (Non-Patent Document 3). In the coherent optical field receiving system, as shown in FIG. 2(A), the optical multilevel signal 123 transmitted over the fiber link is split into horizontal (P) polarization component 133 and vertical (S) polarization component 134 by means of a polarization splitter 131 and then input to coherent optical field receivers 135-1 and 135-2, respectively.
The coherent optical field receiving system should have a local laser 130 having approximately the same wavelength as a transmitting laser. An output light (local light) 132 from the laser 130 is split into two local lights 132-1 and 132-2 by an optical splitter 102 and then input to the coherent optical field receivers 135-1 and 135-2, respectively.
The coherent optical field receiver 135-1 includes an optical phase diversity circuit 136 and a digital signal processor 141. The optical phase diversity circuit 136 generates an I (in-phase) component output light 137 and a Q (quadrature-phase) component output light 138 from a local light 132-1 and a P polarization component 133 of the input optical multilevel signal. The I (in-phase) component output light 137 is in-phase component between the local light and the optical multilevel signal. The Q (quadrature-phase) component output light 138 is quadrature-phase component between the local light and the optical multilevel signal. The I component output light 137 is supplied to a balanced optical receiver 105-1 and the Q component output light 138 is supplied to a balanced optical receiver 105-2. Analog electric signals output from the balanced optical receivers 105-1 and 105-2 are time-sampled by A/D converters 106-1 and 106-2, respectively, and then converted into digital signals.
In the following description, as shown in FIG. 1(E), the optical field of a received signal is defined as r(n)exp(φ(n)) and the optical field of the local lights 132-1 and 132-2 is marked by exp(−θ(n)). Here, r represents the amplitude of the optical field, φ represents the phase of the optical field, and n represents sampling timing and it is assumed that the amplitude of the local light 132 is a constant value “1”. Further, θ(n) represents phase fluctuation which is generated by phase noise inherently included in the laser or by the difference of optical frequency between the local light and the signal light. Although the transmitting laser of a transmitter side also has the phase noise, the phase noise is disregarded for simplification in the following explanation.
Each of the balanced optical receivers 105-1 and 105-2 performs homodyne detection of the input optical multilevel signal with the local light and outputs the in-phase component and the quadrature-phase component of the optical field of the input optical multilevel signal on the basis of the local light. As a result, the output signal 140-1 of the A/D converter 106-1 becomes I′ (n)=r(n)cos(φ′(n)) and the output signal 140-2 of the A/D converter 106-2 becomes Q′(n)=r(n)sin(φ′(n)). For simplification, it is assumed here that φ′(n)=φ(n)+θ(n) and all the constants, such as conversion efficiency and the like, are “1”.
If the phase fluctuation θ(n) is disregarded, φ′(n)=φ(n). As a result, in the case of using the coherent optical field receiver, because all of the information (I and Q components in this case), which represents the optical field r(n)exp(φ(n)), is directly and simply obtained from the received optical multilevel signal 123, optical multilevel signal receiving should be possible. However, the influence of phase fluctuation θ(n) of the local light 132 can not actually be disregarded.
It is assumed, for example, that the received optical multilevel signal is multilevel-modulated in the 16-level quadrature amplitude modulation (16QAM) as shown in FIG. 1(C). When the phase fluctuation θ(n) occurs, the signal constellation of the received signal rotates by θ(n) from an ideal position as equivalently shown in FIG. 2(B). Consequently, it becomes unable to decide which symbol (signal point) is transmitted based on the foregoing I′(n) and Q′(n).
The digital signal processor 141 detects the slow rotation components (˜several 100 MHz) of the signal point from the output signals of the A/D converters 106-1 and 106-2, eliminates the rotation components from the output signals of the A/D converters, assuming the rotation components as the phase fluctuation θ(n), by signal processing, and outputs to a symbol decision circuit 143 output signals 142-1 and 142-2 that represents the correct in-phase component I(n)=r(n)cos(φ(n)) and quadrature-phase component Q(n)=r(n)sin(φ(n)).
The balanced optical receiver 105-2 performs the same operation as the balanced optical receiver 105-1, such that it outputs the correct in-phase component I(n)=r(n)cos(φ(n)) and quadrature-phase component Q(n)=r(n)sin(φ(n)) as the output signals 142-3 and 142-4. The symbol decision circuit 143 judges with high accuracy which symbol is transmitted by comparing the I and Q components output from each digital signal processor 141 with the signal constellation shown in FIG. 1C and output a reconstructed multilevel digital signal 144.
By using the foregoing coherent optical field receiver, it is able to generate all the field information required to decide the multilevel signal by compensating the degradation of the signal due to chromatic dispersion, etc., by the signal processing. Accordingly, in principle, the coherent optical field receiver can receive the complex multilevel signal. Further, the coherent optical field receiver has advantages in that linear degradation due to chromatic dispersion, etc., can be perfectly compensated logically by performing a correction processing on the input signal in accordance with an inverse function of a transfer function of the fiber link by the digital signal processor 141, and there are no restrictions on the compensation quantity. However, since the small and high-speed digital signal processor 141 having signal processing performance of 10 Gbit/s or more has not yet launched onto the market, the foregoing digital processing type coherent optical field receiver is still at the stage where offline processing is performed with a computer on the electric signals 140-1 and 1402-2 obtained by using high-speed A/D converters to verifies the results.
Meanwhile, FIG. 3(B) shows a configuration of the optical multilevel signal receiver for receiving the amplitude and phase modulation light, as disclosed by Non-Patent Document 2. FIG. 3(A) shows an example of an 8-level amplitude and phase modulation light (8APSK) where 8 signal points having quaternary phase and binary amplitude are arranged on a concentric circle. In the optical modulation where the phase components are equidistantly split like 8APSK signals, a differential coding is generally used for modulating the phase components. In the present example, each symbol transmits 3-bit information by correlating each symbol to a binary value amplitude and a quaternary value phase in which phase difference with its just previous symbol is any one of 0, π/2, π, −π/2.
The optical multilevel signal receiver, which receives the 8APSK signal, uses an incoherent scheme that does not detect the optical field and as shown in FIG. 3(B), an input optical APSK signal 124 is branched into 3 optical signals by an optical branching circuit 150.
Among them, two optical signals are input to optical delayed detectors 104-1 and 104-2 and the remaining one optical signal is input to an optical intensity receiver 151. Each of the optical delayed detectors 104-1 and 104-2 includes a first optical path that generates a delay of a symbol time T to the input signal and a second optical path that has a −π/4 optical phase shifter or a +π/4 optical phase shifter and converts the phase modulation components into the optical intensity signals by interfering a state (symbol) of a received optical signal with a symbol received previously by time T.
The intensity of light output from the optical delayed detector 104-1 having the +π/4 optical phase shifter is large when the phase difference between a received symbol and a symbol just before the symbol is 0 or +π/2 and is small when the phase difference between a received symbol and a symbol just before the symbol is −π/2 or π. The output light of the optical delayed detector 104-1 is received by the balanced optical receiver 105-1 and the output of the balanced optical receiver 105-1 is binary-decided by a binary decision circuit 152-1, making it possible to obtain a binary reconstructed digital signal 153-2 corresponding to 1 bit.
The intensity of light output from the optical delayed detector 104-2 having the −π/4 optical phase shifter is large when the phase difference between a received symbol and a symbol just before the symbol is 0 or −π/2 and is small when the phase difference between a received symbol and a symbol just before the symbol is π/2 or π. The output light of the optical delayed detector 104-2 is input to a binary decision circuit 152-2 through the balanced optical receiver 105-2, such that a binary reconstructed digital signal 153-2 corresponding to another 1 bit included in the phase component is reconstructed.
The optical intensity receiver 151 converts the optical intensity (a square of optical field amplitude) of the received signal into the electric signal. The output of the optical intensity receiver 151 is decided by the binary decision circuit 152-3, such that a binary reconstructed digital signal 153-3 corresponding to 1 bit included in the optical amplitude component is reconstructed. Since the optical multilevel signal receiver uses the optical delayed detector, there are advantages in that the optical multilevel signal receiver reduces the influence of the phase fluctuation θ(n) and polarization dependency and does not require the local laser, or the like. Also, the optical multilevel signal receiver is able to receive the APSK signal, up to 16 levels.
Further, FIG. 1 of “A. D. Ellis etc. “Receiver-side Electronic Dispersion Compensation using Passive Optical Field Detection for Low Cost 10 Gbit/s 600 km Reach Applications”, Paper OTuE4, Optical Fiber Conference (OFC) 2006 (Non-Patent Document 4) shows a configuration of an optical transmission system that transmits and receives binary intensity modulation light.
A low cost transmitter shown in FIG. 1 here externally modulates an optical signal output from a DFB laser with an information signal of 10 Gbit/s (a bit width of 100 ps) by a Mach-Zehnder type optical intensity modulator (MZM) or an electro absorptive optical modulator (EAM) and outputs the modulated optical signal to an optical fiber link. The modulation light transmitted by the optical fiber link is filtered by a narrowband optical filter and then received by an optical receiver composed of four parts of “A” to “D”. The optical receiver has an effect of electrically compensating dispersion by obtaining an optical intensity (AM component) waveform and an optical frequency change (FM component) waveform and synthesizing them in the receiver.
The part “A” of the receiver comprises an asymmetrical Mach-Zehnder type optical interferometer with a delay amount of T=5 ps and two photo-detectors (output voltage V1 and V2) coupled to two output ports of the optical interferometer. The optical interferometer has a passing characteristic periodical at an optical frequency period of 200 GHz (=1/T) and serves as an optical frequency discriminator by setting the phase of the interferometer to 90°. For this reason, it is able to obtain an intensity waveform (AM waveform VA) of the received light at the part “B” by operating the sum component (V1+V2) of the output signals from two photo-detectors.
On the other hand, a frequency waveform (FM waveform: VF) can be obtained by dividing the operation result of difference component (V1−V2) by an intensity waveform VA. The part “C” reconstructs pseudo optical field waveform in the receiver by intensity-modulating an electric sine wave signal of an intermediate frequency F (F>10 GHz) obtained from a micro wave oscillator with the AM waveform VA by using an AM modulator and modulating it with a FM waveform VF by using an FM modulator. Thereafter, the obtained electric signal of intermediate frequency passes through a microwave dispersive transmission line having dispersion characteristics opposite to that of the optical transmission line in order to compensate for waveform distortion caused by the dispersion (dispersion compensation) and is received by performing square detection.
The publication merely proposes the principle and does not completely verify that such a configuration is realizable actually. However, according to the proposal, it may be possible to compensate for the wavelength dispersion of the optical signal degraded by the chromatic dispersion.
Non-Patent Document 1:
    R. A. Griffin, et. al., “10 Gb/s Optical Differential Quadrature Phase Shift Key (DQPSK) Transmission using GaAs/AlGaAs Integration,” OFC2002, paper PD-FD6, 2003Non-Patent Document 2:    Kenro Sekine, Nobuhiko Kikuchi, Shinya Sasaki, Shigenori Hayase and Chie Hasegawa, “Proposal and Demonstration of 10-Gsymbol/sec 16-ary (40 Gbit/s) Optical Modulation/Demodulation Scheme,” paper We3.4.5, ECOCNon-Patent Document 3:    M. G. Taylor, “Coherent Detection Method Using DSP to Demodulate Signal and for Subsequent Equalization of Propagation Impairments,” paper We4.P.111, ECOC 2003, 2003Non-Patent Document 4:    A. D. Ellis etc, “Receiver-side Electronic Dispersion Compensation Using Passive Optical Field Detection for Low Cost 10 Gbit/s 600 km Reach Applications”, Paper OTuE4, Optical Fiber Conference (OFC) 2006